Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

Connectivism: Education & Artificial Intelligence

2 249 vues

Publié le

This work is based on a scientific paper: Does Artificial Neural Network Support Connectivism’s Assumptions?
Connectivism was presented as a learning theory for the digital age and connectivists claim that recent developments in Artificial Intelligence (AI) and, more specifically, Artificial Neural Network (ANN) support their assumptions of knowledge connectivity. Yet, very little has been done to investigate this brave allegation. Does the advancement in artificial neural network studies support connectivism’s assumptions? And if yes, to what extent? This paper addresses the aforementioned question by tackling the core concepts of ANN and matching them with connectivist's assumptions.

Publié dans : Formation
  • Identifiez-vous pour voir les commentaires

Connectivism: Education & Artificial Intelligence

  1. 1. Connectivism: Education & Artificial Intelligence Alaa AlDahdouh
  2. 2. Research Question • Does the advancement in Artificial Neural Network (ANN) studies support connectivism’s assumptions?
  3. 3. What is Artificial Neural Network (ANN) • Artificial Neural Network (ANN) is a software structure developed based on concepts inspired by biological functions of brain; it aims at creating machines able to learn like human (Goodfellow, Bengio, & Courville, 2016; Nielsen, 2015; Russell & Norvig, 2010).
  4. 4. What is Artificial Neuron? • Artificial Neuron is a node that receives input from preceding neurons and makes a decision to 'fire' to the next neurons.
  5. 5. What is Artificial Neuron? • Suppose that you are a neuron and you want to make a decision to buy a car. In your perspective, car insurance and gas price are more important than parking cost. Therefore, a car would cost you $575 per month. But you set a threshold to buy a car if it would cost you less than $480. Therefore, you make a decision not to buy a car. • Your own perspectives of inputs and the threshold are called weights and bias respectively.
  6. 6. Artificial Neuron & Connectivism • Connectivists often argue that a person is like a neuron and they both live in networks but in different levels (Downes, 2016). • As each neuron has its own bias, each person has his/her own internal judgement system. Siemens (2005, 2006) states that knowledge and learning rests in diversity of options and Aldahdouh et al. (2015) argues that educational systems should foster the learners' diversity, not their similarity.
  7. 7. Artificial Neuron & Connectivism • In order to be able to learn, ANN should move slowly and smoothly from one decision to another. For these reasons, ANN researchers have examined many alternative soft activation functions such as sigmoid . • Connectivism has been criticized for its oversimplification of interaction between nodes as the connection can be either active or inactive (Clarà & Barberà, 2014). However, connectivism proponents (Aldahdouh et al., 2015) have shown that a connection is graded and not necessarily sharp.
  8. 8. What is ANN Architecture? • ANN Architecture refers to the way of arranging neurons in certain order to make it easier for a learning algorithm to find the biases and weights. ANN levels of abstraction ANN as a person’s brain ANN as a group of learners Architecture levels of abstraction Learner's inner abilities and mental capacities Designing learning- environment
  9. 9. What is ANN Architecture? • The most common ANN architectures can be divided based on three criteria: – Number of layers – Flow of information – Neuron connectivity
  10. 10. ANN Architecture: Number of Layers Shallow Neural Network Deep Neural Network
  11. 11. ANN Architecture: Number of Layers • The terms shallow and deep are somehow misleading because they are not in line with educational terminology of surface and deep learning (Vermunt & Vermetten, 2004). • The concept of layers is completely incompatible with connectivism’s assumptions. The idea of that a network consists of a sequence of layers contradicts with chaos theory which is one of the underpinning theories of connectivism (Aldahdouh, 2017).
  12. 12. ANN Architecture: Flow of Information Feedforward neural network Recurrent neural network
  13. 13. ANN Architecture: Flow of Information • Flow of information and connection directionality are some of subjects discussed in connectivism literature. Aldahdouh et al. (2015) showed that some connections in knowledge network are bidirectional while others unidirectional. They also showed that "The node can connect to itself" (p. 5). • Information in connectivism flows in both directions which means feedforward architecture contradicts with connectivism’s principles while recurrent architecture agrees with it.
  14. 14. ANN Architecture: Node Connectivity Fully connected neural network Convolutional neural network
  15. 15. ANN Architecture: Node Connectivity • Connectivism appreciates network connectivity and seeks to increase it as much as possible. Actually, connectivism defines learning as the process of connecting nodes in a network (Aldahdouh et al., 2015; Siemens, 2005). This may indicate that connectivism aims to make a learner as a node in the fully connected network. • However, it has been proved that increasing connectivity adds complexity to ANN. A convolutional network, on the other hand, decreases the connectivity and achieves better results. Connectivists should pay attention to this because it disagrees with their main network designs (Downes, 2010a). • One can argue that connectivism agrees with fully connected network but disagrees with convolutional network.
  16. 16. Learning Algorithm: Training Set • Suppose you have a very simple neuron with one input and one output. To teach this neuron to memorize a multiplication table for number 5, ANN researchers usually give it a so-called training set. • A Training Set contains a number of different input values (1, 2, 3, 4, 5, 6 ...) paired with the correct output (5, 10, 15, 20, 25, 30 ...).
  17. 17. Learning Algorithm: Training Set • The labeled training set which contains the input values along with correct output assumes knowledge as something static and something we know in advance. Learning algorithm is not allowed to manipulate inputs or correct outputs in any case (Nielsen, 2015). • This limits the ability of ANN to learn something previously known, not to discover something new. • The idea of static knowledge contradicts with connectivism’s principle of dynamic knowledge (Aldahdouh et al., 2015).
  18. 18. Learning Algorithm: One Neuron Training • The neuron receives input and generates output according to its own weight (w) and bias (b) which were randomly selected. This means, the output of the neuron (as) would most probably differ from the correct output (ys). Cost If the Cost goes to zero, then we found the right w and b.
  19. 19. Learning Algorithm: One Neuron Training • Assume a neuron = a person => The inputs of the neuron would represent the current learning experiences. The correct outputs represent the reality (ontology) and neuron outputs represent a person's perceptions about the reality (epistemology). • The Cost represents the gap between learner's perceptions and the reality. That is to say, learning is the process of minimizing the gap between learner's perceptions and the reality. • Of course, this definition perfectly fits constructivist theory of learning (Piaget, 2013).
  20. 20. Learning Algorithm: Finding the minimum Cost
  21. 21. Learning Algorithm: Finding the minimum Cost • A learning rate (η) refers to the speed of learning outcome. Or how fast a learner should learn. The learning rate should not be too fast that makes a learner jump from point to point; long jumps disrupt learning. A learning rate should not be very slow too; it makes a learner crawl in detail that would not serve him to achieve his goal. Finding the right pace of learning is a difficult task that depends on the initial state of the learner’s perspectives and bias. • Learning rate is one of many other free parameters which are left free for human and outside of ANN’s control. • Adjusting free parameters is the responsibility of consciousness. • Connectivism is also criticized for its ambiguity in that it does not show how pattern recognition is done (Aldahdouh et al., 2015).
  22. 22. Learning Algorithm: Network Level • In a network of millions of neurons, what are those right weights and bias that a single neuron learns? What is the meaning of those connections and biases? Why does each neuron connect to other neurons in that way? Until now no one has a theory. • Learning in network level clearly supports the core assumption of connectivism: A single connection between two nodes does not have meaning in its own. A meaning is distributed across group of connections or patterns (Aldahdouh et al., 2015; Downes, 2006). Looking at the network from higher level mitigates its complexity (Downes, 2016). But that does not give us the answer and the exact meaning of the entities in the lower level.
  23. 23. Aldahdouh, A. A. (2017). Does Artificial Neural Network support Connectivism’s assumptions? International Journal of Instructional Technology and Distance Learning, 14(3), 3–26. Read More Alaa AlDahdouh